Too often I disagree with the RESULTS of the methodology used to analyze data releases. There is no methodology which is perfect – and for this reason I believe the less you screw with the raw data – the more consistent the results for baselining, comparing, and trending.

This week there was one confusing data release – retail sales – where the headlines continue to show worsening retail sales. Honestly, I really never will understand the seasonal adjusting analysis by the US Census. The unadjusted data clearly shows fairly strong retail sales.

Unadjusted Retail Sales by Month and by Year

And if one looks at percent growth year-over-year – February was a very good month.

Advance Retail Sales Year-over-Year Change – Unadjusted (blue line), Unadjusted with Inflation Adjustment (red line), and 3 Month Rolling Average of Unadjusted (yellow line)

Pay attention to the rolling average (yellow line in the above graph) – it shows improvement. One month of data can be an anomaly – but averaging data over time smooths out the unusual impacts (such as weather). Still, the rolling averages are still below the levels seen in 2014, and the results are not showing good times are here again.

Even when looking at the year-over-year growth of the adjusted headline data and the unadjusted data, both show improvement year-over-year.

Comparison of the Year-over-Year Census Seasonally Adjusted Retail Sales (blue line) and Econintersect’s Unadjusted Retail Sales (red line)

And to to continue to drive home the point that only US Census sees a decline in growth – here is employment data for the retail sector:

Retail Employment – Total Seasonally Adjusted (blue line, left axis) and Year-over-Year Change Unadjusted (red line, right axis)

I would not go to the bank with retail sales data in any event. The data will be adjusted for the next few months (which could adjust away some of the increase). Even though retail sales is very important to an economy based on consumption, this data point is not reliable, especially in real time, UNLESS confirmed by other data points.